Urinary copper is linked to regional cortical volume reductions in adolescents with major depressive disorder

Major depressive disorder (MDD) is highly prevalent among adolescents, but its neurobiological mechanisms remain unclear. Neuroimaging studies have shown cortical abnormalities in MDD, and emerging evidence suggests that trace elements may impact brain structure. This study aimed to investigate the association between urinary element concentrations and cortical alterations in adolescent MDD. Structural magnetic resonance imaging (sMRI) was conducted on 190 adolescents with MDD and 123 healthy controls (HCs), measuring cortical volume, cortical thickness, and urinary metal concentrations. Partial correlation analyses examined the associations between differential elements and cortical features, while mediation analyses tested whether cortical changes mediated the effects of elements on symptom severity. Furthermore, nine supervised machine learning models were trained using differential cortical and element features, and model performance was evaluated using cross-validation and stacking ensembles. The results revealed significant reductions in cortical volume in the left lateral orbitofrontal cortex (LLOF), left medial orbitofrontal cortex (LMOF), left rostral middle frontal gyrus (LRMFG), and left frontal pole (LFP), as well as reduced cortical thickness in the right pars triangularis (RPTRI), right pars orbitalis (RPORB), and right medial orbitofrontal cortex (RMOF) in adolescents with MDD compared with HCs. Urinary copper (Cu) levels were negatively correlated with LFP volume and positively associated with both depressive and anxiety symptom severity. In machine learning classification, the stacking ensemble demonstrated the best overall performance, indicating the discriminative value of combined cortical and urinary element features in distinguishing adolescents with MDD from HCs. This study suggests that trace elements, particularly Cu, are linked to structural brain alterations and symptom severity in adolescent MDD. The integration of neuroimaging and element data enhances predictive modeling and offers new insights into the pathophysiology of MDD.